Skip to content
← Back to home
ESENCambiar a español

UTEC · Java EE Information Systems Workshop

Electric Mobility Management

A system to manage electric-vehicle charging, built with Jakarta EE on WildFly.

Role
Team development: payments, mobile-app authentication and LLM messaging
Year
2026
Stack
Java 17Jakarta EE 10WildFlyMariaDBJMSGrafanaOllama

The problem

Java EE workshop: a system to manage electric-vehicle charging across distributed stations. Customers start and finish charges, manage their payment methods and file complaints. It was built in four iterations, each adding a new problem: integrating external systems, observing the system in production and decoupling slow processes.

What I built

  • Integration with a payment-method system and with UTE's billing, both simulated with mocks, including handling of rejected payments and outstanding debt.
  • Authentication and authorization for the mobile app, and REST APIs for the app and the web manager.
  • Observability: charge and payment metrics with Micrometer, stored in InfluxDB and shown on a Grafana dashboard.
  • Messaging: a JMS queue for complaints, processed asynchronously by a local language model (Ollama with Llama 3.2) that classifies them.
  • A token-bucket rate limiter on the charge history, tested with a JMeter plan, and Postman collections to exercise the whole flow.

Technical decisions

A modular monolith that could be split

Modules only talk through the interfaces of their application layer, so any of them could become a microservice without rewriting its callers. It gives the order of separate services without paying the cost of operating them separately.

A queue between the complaint and the LLM

Classifying a complaint with a local model takes time. With a point-to-point queue the customer gets an answer right away and classification happens afterwards, even if the model is slow or down.

Measure before something fails

Metrics for active charges, completed charges and approved or rejected payments were defined together with the dashboard, to watch how the system behaves and not just find out when something breaks.

What I took away

  • How a Jakarta EE application works inside: CDI, JPA, REST, security and messaging on an application server.
  • Integrating external systems that can fail or reject payments, and modelling those cases instead of ignoring them.
  • Instrumenting a system with metrics and reading them in Grafana, and protecting an endpoint with a rate limiter.